WhatsApp Database Segmentation: Don't Send Everyone the Same Thing
AndySendy academy
← All posts

🔗 Same text to the whole database isn't saving time - it's burning contacts

Sending identical messages to those who replied, those who read and stayed silent, and those who never received it - common when rushing. Result is predictable: happy clients get annoyed by repeats, silent ones report spam. How WhatsApp technical statuses become real segments and why Excel is a logging tool, not a data source.


Why one text for the whole base is a mistake, not savings

WhatsApp returns four technical message statuses: sent (reached server), delivered (reached device), read (opened by recipient), and failed (delivery error). Real system events underpinning automatic database tagging in CRM and mailing services - if the base was built correctly from the start, as in incoming leads breakdown.

Each status needs a different next step, not the same message on repeat.


Before / After: where the original thesis needs correction

Before:

Different messages per segment boost response 2–3x.

After:

Per agency cases, customized offer for "read but no reply" segment - objection handling or bonus - raises conversion 2–3x vs resending same text to entire base. No official Meta confirmation - practical benchmark, not documented standard.

Before:

How to tag the base directly in Excel.

After:

Excel is a static table - it doesn't receive delivery/read statuses on its own. Dynamic segmentation needs CRM or WABA API catching Meta webhooks; Excel gets exported logs for filtering and analytics.

Additional original inaccuracy:

"Delivered" = "didn't read."

Wrong. If recipient disabled read receipts in privacy settings, read never forms even after real reading - message stays delivered forever. Part of active audience lands in "didn't read" segment due to privacy settings, not real ignoring.


Four statuses - four different scenarios

Status Technical meaning Next step
Read, no reply Opened, no inbound from JID within timeout Handle objection, bonus, clarifying question - not same text repeat
Delivered, no read Reached device, read status not formed (including disabled receipts) Duplicate channel (SMS) or soft reminder, no aggression
Failed Usually number not registered on WhatsApp Remove from base - resend wastes limits
Replied Two-way dialog exists Exclude from bulk blast - continue only in dialog context

Not three segments from original thesis - minimum four; failed deserves separate handling.


How to tag technically: API to Excel

Dynamic segmentation starts not in Excel but in a system catching statuses:

In all three cases Excel stores and filters received statuses - not the collection source.


Mini-case: VLOOKUP tagging vs carpet bombing

E-commerce marketer exported first wave report to CSV, VLOOKUP-filtered in Excel: failed removed as invalid, read no reply tagged "Segment B" with 10% discount reminder, delivered no read - "Segment C" with SMS duplicate offer. Repeat-touch revenue up 140%, no blocks recorded.

Shows: splitting base on at least two axes (read/didn't read) with different messages per axis gives measurable gain without harming number. Doesn't prove: 140% reproducible in any niche - one documented result.

Opposite: beauty salon sent promo to 3000-number base, same text again 3 days later to entire base including those who booked or clearly ignored. Repeat perceived as spam - mass complaints, gray work number banned at 200 sends in wave two. Difference - segmentation before repeat touch, not technology - see mass mailing ban mechanics.


Debated point: when to follow up "read but no reply"

Practitioners disagree on timing. Some say follow-up 2–4 hours after read while attention still on chat. Others argue that speed looks like bot stalking and triggers complaints - safe interval strictly 24–48 hours in their view.

No confirmation either way - practical preference, not documented rule - like behavioral antispam guides.


Numbers to keep in focus

Confirmed activity segmentation guides: active - opened in last month; inactive - no activity 3+ months; "last read within 60 days" - practice for contacts who opened chat in last 2 months. Excel integration via ready services - ~5 minutes setup with up to 30-day trial.

Forum, undocumented Meta metrics: 2–3x conversion lift personalizing "read no reply" vs same text to whole base; 15–25% users disable blue read ticks, systematic error in manual segment export; behavioral slices reduce complaints 40–60% per practitioner observations.


Common misconceptions

"Delivered means definitely didn't read" Wrong. Read via notification shade, preview mode, or disabled read receipts - all stay delivered.

"Excel tracks reads in real time" Wrong. Static table; dynamic read/delivered segmentation needs webhook-catching system; Excel gets final log only.

"No reply = send same text again" Wrong. Repeat without context change = spam - beauty salon case proves it.

"Failed = try resending tomorrow" Wrong. Usually number not on WhatsApp - resend wastes limits.

"Read no reply = not interested" Wrong as universal rule. Busy, deferred decision - objection-handling segment, not write-off.


Practical takeaways


🎯 Next step

If mailings go one text to whole base now - start simple: export last campaign statuses, split at least read/didn't read. Already reduces re-irritating those who saw message and didn't reply.

Conclusion

Practical rule:

One text for the whole base is a message not to the client but to yourself: "I can't be bothered to see who reacted how."